***************************************************************** * * Titel: Barbiero's Reply Autor: Daniel Barbiero, Silver Spring - USA Dateiname: 13-2-96.TXT Dateilänge: 55 KB Erschienen in: Wittgenstein Studies 2/96, Datei: 13-2-96.TXT; hrsg. von K.-O. Apel, N. Garver, B. McGuinness, P. Hacker, R. Haller, W. Lütterfelds, G. Meggle, C. Nyíri, K. Puhl, R. Raatzsch, T. Rentsch, J.G.F. Rothhaupt, J. Schulte, U. Steinvorth, P. Stekeler-Weithofer, W. Vossenkuhl, (3 1/2'' Diskette) ISSN 0943-5727. * * ***************************************************************** * * * (c) 1996 Deutsche Ludwig Wittgenstein Gesellschaft e.V. * * Alle Rechte vorbehalten / All Rights Reserved * * * * Kein Bestandteil dieser Datei darf ganz oder teilweise * * vervielfältigt, in einem Abfragesystem gespeichert, * * gesendet oder in irgendeine Sprache übersetzt werden in * * irgendeiner Form, sei es auf elektronische, mechanische, * * magnetische, optische, handschriftliche oder andere Art * * und Weise, ohne vorhergehende schriftliche Zustimmung * * der DEUTSCHEN LUDWIG WITTGENSTEIN GESELLSCHAFT e.V. * * Dateien und Auszüge, die der Benutzer für * * seine privaten wissenschaftlichen Zwecke benutzt, sind * * von dieser Regelung ausgenommen. * * * * No part of this file may be reproduced, stored * * in a retrieval system, transmitted or translated into * * any other language in whole or in part, in any form or * * by any means, whether it be in electronical, mechanical, * * magnetic, optical, manual or otherwise, without prior * * written consent of the DEUTSCHE LUDWIG WITTGENSTEIN * * GESELLSCHAFT e.V. Those articles and excerpts from * * articles which the subscriber wishes to use for his own * * private academic purposes are excluded from this * * restrictions. * * * ***************************************************************** Abstract: With this Reply, I offer further remarks on explanatory adequacy and the problem of reproduction. 1. THE SKEPTICAL SOLUTION AS EXPLANATION [1] Is it legitimate to raise the question of the explanatory adequacy of Kripke's skeptical solution? On the surface, it would appear not. The Wittgensteinian position generally is that the search for explanation is philosophically misguided, and one can hardly speak of explanatory adequacy if there is no explanation to speak of. But it seems to me that Kripke's account does in fact function as an explanation. [2] To provide an explanation, in the sense in which the term will be used here, is to show how a phenomenon can be understood in terms of its relevant constituent factors. I would distinguish this kind of explanation, which for present purposes might provisionally be termed a constitutive explanation, from a strictly causal explanation. In a case in which, for example, a response or set of responses is observed of an individual or in a selected population, a constitutive explanation would provide us with some account of the constituent factors in relation to which that response or set of responses is constituted as such. In a sense, a constitutive explanation redescribes the phenomenon to be explained at a level different (perhaps "deeper" is a word that will fit here) from that at which we describe it before we attempt to explain it. Note that, again, this kind of explanation need not entail reference to causal factors of any sort. [3] It should be clear that as it is understood here, explanation has nothing to do with any (entirely spurious) attempts to, as Huen puts it, "uncover the veiled essence of life" (Huen b 2.2). At best, such attempts may yield pseudo-explanations of some sort -- that is, statements that may have the form of explanations, but which in fact explain nothing. If in raising this possibility it was Huen's purpose to present an either/or choice between a Wittgensteinian approach in which one is invited only "to acknowledge that life is so going on" (Huen b 2.2) and the kind of pseudo-explanation that purports to reduce things to their supposed "essences," then clearly he is offering a false choice. This becomes even clearer when we consider that, for reasons to be shown below, Kripke's skeptical solution is, for all intents and purposes, itself an explanation of the type described above. [4] Huen, it should be remarked, indirectly acknowledges the explanatory drift of Kripke's argument. For as he says in regard to the skeptical solution's claim that it is a brute fact that people in a community tend to agree in their responses, "this brute fact [of agreement] may be said to be an `explanation' in the sense that by reference to it, the puzzle about the possibility of rule-following is settled" (Huen b 2.2). This is precisely the issue, and I see no reason to put scare quotes around the word explanation. Solving such a puzzle, if we would like to call it that, is precisely what an explanation, in the sense discussed above, should do. Accordingly, I want first of all to show how, in the most general sense, Kripke's skeptical solution counts as an explanation of a given type of behavior. Following that, I will show how the skeptical solution provides a specifically constitutive explanation of that behavior. 1.1. The Skeptical Solution as Covert Explanation [5] Kripke claims that the skeptical solution purports only to say something about the conditions under which certain assertions can be made regarding apparent rule-following behavior and, more generally, concept possession. On the surface, at least, this does not look much like a claim for the explanation of that behavior. But the surface here is deceiving. [6] It seems fairly clear that when, under certain circumstances, we claim that certain things can be asserted about someone's behavior, we are at the very least licensing a given explanation of that behavior. To say, for example, that we are justified in asserting X of Jones's behavior is to say that we are justified in using X to describe or explain Jones's behavior, depending on the context, the content of X, our purposes in asserting X, and so forth. In other words, under the proper circumstances, a justified assertion may count as a valid explanation. If, in the case of Jones's apparently following the rules of addition, we claim that the only justified assertion we can make regarding his behavior is that his responses agree with those likely to be given by members of his community in the appropriate circumstances, then we are in fact explaining Jones's apparent rule-following behavior in terms of his agreement or disagreement with the responses of the appropriate members of his community. [7] What I am getting at here is that while Kripke's skeptical solution seems only to be about certain moves in a language game rather than about the explanation of behavior per se, the language game in question is the one in which we attempt to explain behavior. We might say in fact that it is the language game of explanation within the framework of folk psychology -- ignoring, for present purposes, the negative connotations this latter designation may carry. Such kinds of explanatory conjectures, while lacking the rigor and formality of scientific theories of behavior, do serve us -- often very well -- in our everyday efforts to explain and predict the behaviors of others. Accordingly, Kripke's skeptical solution (like much of Wittgenstein's later work, for that matter) can be understood as a critique -- and to a large extent an intended correction -- of certain explanatory strategies encountered in folk psychology. [8] Thus when Kripke states that Wittgenstein thought that "all talk of an individual following rules has reference to him as a member of the community" (WRPL 109), he seems to be saying that when we attempt to explain someone's behavior by saying that he or she is following a rule, what we're really saying is that his or her responses agree with those likely to be given by members of his or her community. This latter would then count as the actual explanation for the behavior under consideration. As I suggested above, it seems to me that in taking this line, Kripke seeks to correct a commonly offered folk-psychological explanation of a given type of behavior (i.e., that the person is correctly following a rule) by replacing it with another (i.e., that the person's responses agree with those of his or her community). The skeptical solution thus compels the conclusion that explanations in terms of individual rule-following do not in fact explain behavior, since (as it is claimed) reference to rule-following turns out to refer to no fact about the supposed rule-follower, but rather to the "brute fact" that members of a community agree in their judgments. The upshot is that -- again, in this context -- replacing talk of rule-following with talk of agreement is, in effect, replacing one explanation with another. (Note, though, that when we claim that one explanatory account is supposed to correct another we are not necessarily committing ourselves to saying that one account is true and the other false. I will have more to say about this later.) [9] In sum, the skeptical solution offers a de facto explanation of behavior by virtue of its purported correction of an allegedly mistaken move in the folk psychological language game of behavioral explanation. It is for this reason that we might characterize it as a covert, rather than an overt, explanation of apparent rule-following behavior. 1.2. The Skeptical Solution and the Structure of Explanation [10] It is interesting to note that Kripke's replacement of talk of rule-following with talk of agreement fits the pattern of a classic explanatory reduction. Consider that Kripke presents rule attribution as part of a broader, more general strategy by which we ascribe concept possession to people. By subsuming the particular phenomenon (rule-attribution) under a more general, covering phenomenon (concept attribution), Kripke is effectively making the case that a range of phenomena broader than that under specific consideration can be reduced, for purposes of explanation, to a single principle broad enough to account for the general class of relevant cases. The single principle to which both the specific phenomenon (ascription of rule-following) and the general phenomenon of which it is a subtype (ascription of concept possession) are reduced is, of course, agreement. [11] It is significant in this respect that Kripke formulates this principle in classic explanatory form -- that is, as a lawlike conditional, albeit a contraposed one (WRPL 95). The structure of Kripke's formulation serves to trace the explanatory link between the explanans (agreement) and the explanandum (attribution of concept possession) in such a way that we are allowed to see the explanandum as situated within a range of possibilities dependent on whether or not the conditions set out in the antecedent obtain. Further, this treatment allows us to see the explanandum as non-unique in relation to occasion and this, as Pylyshyn points out (Kasher 242 and 250 n 10) is precisely what we would want from an explanatory theory. 1.3. The Skeptical Solution as Constitutive Explanation [12] As noted above, when we provide a constitutive explanation for a phenomenon, we redescribe it in terms of its relevant constituent factor or factors. I think it should be clear that, given the skeptical solution's argument, the relevant constituent factor involved in the attribution of rule-following is understood to be the agreement between the rule-follower's responses and the responses of his or her community, in enough particular cases. Accordingly, Kripke redescribes Jones's apparent rule-following behavior as consisting in Jones's agreement, in enough particular cases, with the responses likely to be given by members of his community. [13] Consistent with its position that agreement is the factor constituting (apparent) rule-following behavior as such, the skeptical solution offers agreement as a brute fact. I believe we are justified in interpreting "brute fact" in the usual sense, that is, as a fact that obtains without any other fact obtaining. Given the context of Kripke's argument a brute fact would be a fact, relative to the folk-psychological language game in which we explain Jones's apparent rule-following behavior, to which we can appeal in explaining Jones's behavior, and which we need not account for in terms of any other facts. It is for this reason that I referred to it in my earlier paper as an explanatory primitive (Barbiero 1996 section 2.2). If agreement is the brute fact in terms of which Jones's responses are meaningful, then there is nothing more to be said in explanation of Jones's apparent rule-following other than that his responses agree, in enough instances, with those likely to be given by others in his community. Explanation must indeed come to an end somewhere, and for Kripke, the explanation of apparent rule-following behavior comes to an end with the invocation of agreement. 1.4. Agreement and the Relevance of Individual Cases [14] In my earlier paper, I framed the matter of agreement in terms of particular cases and the responses of individual members of the community. Huen has objected to this (Huen b 2.4). But it seems to me that a consideration of agreement from the point of view of individual cases is justified for two reasons, one of them general, and the other specific to Kripke's essay. [15] As a general matter, social accounts of meaning -- and Kripke's certainly is one -- carry a certain burden of proof. (I note that this is an observation Chomsky makes (Kasher 31).) It is the burden of social theories to provide a convincing, or at least plausible, account of how individuals within the given population share meanings they are said to share. The point is that a social account of meaning must be able to show, in principle, how individuals conform to the regularities that are supposed to function as, constitute, or undergird meaning, depending on the claim of the specific social theory under consideration. It seems to me that if we are to grant social theories of meaning any explanatory validity, we must do so on the basis of their carrying this burden adequately. [16] More specific to the current discussion, Kripke himself frames the matter of agreement in terms of particular cases. In fact the effect of his discussion is to present agreement as a kind of statistical matter emerging from an aggregation of individual instances. Consider the terms in which he describes the standard communities are supposed to use when ascribing concept mastery to any of their members: "Any individual who claims to have mastered [a] concept...will be judged by the community to have done so if his particular responses agree with the community in enough cases" (WRPL 91-92). [17] Simply put, the standard consists in the condition that the given person come up with the appropriate response often enough to merit attribution of concept mastery. (Perhaps we can say that in enough cases of agreement, agreement is statistically significant enough to merit our saying that the person in question really "has got it.") [18] Kripke is speaking here specifically of the concept of addition, but the generalization I have made is consistent with those he makes elsewhere. The point here is that individual instances count. I need not belabor the fact that the phrases "particular responses" and "enough cases" allow no mistaking that we are in fact considering agreement in terms of individual cases, and Kripke's reference to "[a]ny individual" is self- explanatory. Additionally, one could point to instances in which Kripke refers to "particular inclinations to give particular answers to particular addition problems" (WRPL 91), licenses the attribution of concept possession to someone on condition that that person's "answers to particular...problems agree" with others' (WRPL 91), or, in describing a teacher's grounds for attributing mastery of addition to a child, holds that the child must get a "certain number" of problems correct (WRPL 90). (I am taking "certain number" in the only way it makes sense in this passage, that is, as a significant sum arising from an aggregation of specific instances.) Should we ignore the significance of the language I have (admittedly inexhaustively) cited above, we need only heed Kripke's reminder that "the point is that if, in enough concrete cases, Jones's inclinations agree with Smith's, Smith will judge that Jones is indeed following the rule" (WRPL 91). Clearly, even if Kripke rejects facts about individuals as a legitimate basis for explaining their behavior, he does acknowledge the importance of individual cases and particular instances as factors to consider when ascribing concept mastery. 1.5. Explanatory Constituents, Agreement, and Forms of Life [19] It would be useful here to look briefly at the relationship Kripke sets up between agreement and the Wittgensteinian concept of the form of life. For however it may function in Wittgenstein's later thought, form of life for Kripke tends to be presented in terms of agreement. This reinforces the idea that it is the latter notion, not the former, that provides the relevant explanatory factor (in the sense discussed above) of apparent rule-following behavior. [20] Although Kripke devotes a little less than two pages to a sustained discussion of the notion of the form of life, a clear picture of the relationship between agreement and form of life does emerge. In essence, Kripke presents the notion of form of life in such a way as to establish the phenomenon of agreement as fundamental to the existence of the form of life, as indeed comes out clearly in his definition of a form of life in terms of its presumably underlying agreement(s)*1*, specifically, as a "set of responses in which we agree, and the way they interweave with our activities" (WRPL 96). Kripke can be read here as conceding the organic/holistic drift of some of Wittgenstein's remarks on forms of life, but the emphasis clearly is on the agreement of responses, which in this formulation (as indeed is the case throughout the course of Kripke's discussion of forms of life in WRPL 96-98) is effectively presented as playing a constitutive role vis-a-vis the form of life. It is difficult to avoid the conclusion that as far as Kripke's skeptical solution is concerned, forms of life are established on the basis of a constituent set of agreeing responses. In short, for Kripke, the form of life would appear to depend on*2* the relevant agreement of response rather than vice versa. [21] (Tellingly, the notion of form of life takes second place to agreement in Kripke's ordering and explication of the skeptical solution's three key concepts (WRPL 96-99). Indeed, in his summing up of the essay's argument (WRPL 107-109) he makes no mention of the concept of form of life; while the summary's allusions to agreement and community, especially at point 6 (WRPL 109), may be interpreted as implying the latter concept, it is significant that it is precisely in terms of agreement -- and only in terms of agreement -- that such an implication can be extracted, and with a bit of effort at that.) 1.6. Kripke's Wittgenstein, or Kripke's Kripke? [22] The concept of form of life aside, it seems to me that it is important to appreciate fully the extent to which Kripke, while purporting to expound on Wittgenstein, is in fact doing something distinctively his own. As noted above, Kripke's skeptical solution can be read as a critique of certain folk-psychological explanatory strategies, and in this sense it is consistent with much of Wittgenstein's later work. But whereas Wittgenstein's approach generally is not to explain but simply to point out examples illustrating how or that such and such is done, Kripke does something very different: he analyzes, formalizes, and ultimately, explains. As he acknowledges, he expresses what he takes to be Wittgenstein's views "more straightforwardly" than Wittgenstein would have done himself (WRPL 69). As I have shown above, the resulting product is a generalized, reductive, explanatory account -- something far different from Wittgenstein's deliberately unsystematic remarks. [23] In addition, Kripke does all this in a way that is consistent with a framework that he established for himself over the course of his own work. For it seems to me that there is a striking resemblance between the theory of meaning embodied in the skeptical solution and the so-called causal theory of reference central to Kripke's earlier work on naming. If the former insists that assertions are justified in relation to agreement within a given community, the latter establishes naming on the basis of a series of referring events that depend on what is, in effect, agreement within a referring community to use the name in the given way. This resemblance holds even when we take into account the intentionalism contained in the earlier theory. It would seem then that with Kripke's Wittgenstein, one gets as much Kripke as Wittgenstein. [24] Or perhaps more Kripke than Wittgenstein. Kripke's departures from Wittgenstein's line of thought in the private language argument have been well-documented. Stern, for example, has convincingly shown that Kripke's skeptical solution virtually inverts Wittgenstein's thinking on language and meaning, and offers a highly vulnerable, consensus-based solution where Wittgenstein offers a "straight solution" (Stern 176-181.) Wright has documented similar discrepancies between Kripke's and Wittgenstein's framing of the private language argument (Wright 242-243). Fogelin has shown that Kripke's reliance on public checkability depends on a selective deployment of skepticism, and that it is itself liable to being undermined by the skeptic's all-encompassing doubt (Fogelin 245). As Fogelin further notes, Kripke in a belated footnote acknowledges, but does not explore the implications of, this possibility (Fogelin 246 on WRPL note 87 added in proof). In sum, it is difficult to avoid the conclusion that a close identification of Kripke's skeptical solution with Wittgenstein's own views -- which is what I take Huen to be arguing -- should leave us, well, skeptical. Even so, the questions Kripke raises, whether they are considered a novel interpretation or a misinterpretation of Wittgenstein, are interesting in their own right. 1.7. Explanatory Adequacy [25] As I have argued, the skeptical solution reduces a given folk-psychological explanation of concept possession to the purportedly brute fact of agreement. Does this in fact provide an adequate explanation for the phenomenon in question? I argued in my earlier paper that it does not. What I would like to do now is explain briefly what I mean by explanatory adequacy. [26] Explanatory adequacy is something Chomsky has insisted on since the generative linguistics program was first formulated. Following Chomsky, we can say that a theory provides an adequate explanation if it allows all facts relevant to its domain to be derived from the entities or factors it posits to account for those facts. Further, the explanatory theory must be such that the set of admissible hypotheses will be minimal (KL 55). All told, an adequate explanation will provide comprehensive coverage of all relevant data while using the minimum necessary explanatory factors. While accepting Chomsky's requirements of comprehensiveness and parsimony, I would like to make explicit a standard that those requirements may be seen to imply: that of accuracy. [27] For present purposes we might say that there are explanations of greater or lesser explanatory reach, or accuracy, and that the former are to be preferred to the latter. An explanation can thus be understood as an assertion or set of assertions that may be more or less accurate, as the case may be, rather than simply true or false. Adequacy here therefore need not be defined in terms of whether or not a given conjecture is true or false or meets given necessary and sufficient conditions, but rather in terms of whether or not it provides for an explanatory model that is more accurate and/or accounts for a more significant range of the data than a competitor. It might therefore be useful to think of explanations as locations within a graded space of approximation. If this is the case, then we might accept one explanation over another by virtue of the fact that it more comprehensively tells us what we need to know compared to its competitor, or is the closer or more plausible approximation, given the kind of phenomenon to be explained, the context in which explanation is called for, and so forth. [28] It follows that talk of one explanation being offered in correction of another does not necessarily commit us to dealing in truth conditions, and consequently to holding that the preferred explanation is true and the rejected one false. What explanatory correction does commit us to in practical terms is the position that the preferred explanation refers to a constituent element or elements that are of an order appropriate to account for the broadest range of relevant phenomena and to carry sufficient explanatory force across relevant cases. What we need to fulfill this commitment is an explanatory agent that is of the right order to function as a plausible constituent factor while not, remaining within the relevant context, itself begging explanation. [29] I argued in my earlier paper that agreement is not such a factor, and that we have to say more than that a person's responses agree with those of his or her community. Specifically, it is still left to us to explain how it is that his or her answers can come to agree with those of the community. This the skeptical solution does not do. Let me make clear here that my objection is not to Kripke's effective approach, which is to explain, by default or design, a given phenomenon, but rather to the termination of the chain of explanation at the ostensibly brute fact of agreement. My criticism rests on the feeling that agreement is not the place to stop; agreement of response, in other words, is of the wrong order to be considered brute. Agreement itself is what needs to be explained -- a position, interestingly, that Wright suggests in his perceptive essay on the Kripke-Chomsky controversy (Wright 245). [30] This leaves us with a question: How can publicly observable norms produce the behavioral regularities they apparently do? How does a community come to agree in its judgments, in other words? This brings us to the problem of reproduction, and it is to that problem that I turn next. 2. REPRODUCTION REVISITED [31] In my earlier paper, I focused much of my criticism of Kripke's skeptical solution on its failure to address adequately what I called the problem of reproduction. I argued that although a community's behavioral regularities may indeed provide normative examples for its individual members to follow, an adequate explanation of how these examples carry the normative force we attribute to them must somehow account in turn for how they can be reproduced by members of the community. Accordingly, I sketched a simple model of a process of learning in order to show one form such an explanation might take (Barbiero 1996 section 5). Here I would like to amplify some of the comments I made, and perhaps throw into sharper relief some features that were not emphasized in the earlier paper.*3* 2.1. The Problem of Reproduction [32] The problem of reproduction is the problem of how individuals within a designated population come to reproduce the behavioral regularities held to be characteristic of that population. As I put it in my earlier paper (Barbiero 1996 section 4), the problem of reproduction comes down to this: How can we account for the fact that members of a group seem to instantiate behaviors that, taken in the aggregate, exhibit a certain consistency or regularity? How can it be that the behavior exhibited by community member A in a given situation is also reasonably approximated by community member B in a relevantly similar situation? How is it, in other words, that we can get individuals within a designated community to agree, and thus to be eligible to be considered a part of that community, in the first place? In specific regard to the skeptical solution, the problem of reproduction is the problem of explaining how different individuals can in fact come to the agreement that is supposed to constitute their (apparent) rule-following behavior as such. 2.2. Reproduction and Learning [33] It seems to me that if we are to show how members of a community can come to agree in their responses, we must come up with some sort of account of learning. As it happens, I think a consideration of what is involved in learning renders implausible the appeal to agreement as brute fact. Consider that if we answer the question: What is it that licenses us to say that Jones grasps the concept of addition? with this: Because he has been appropriately trained we now legitimately can be asked how it is that Jones's training allows him to grasp the concept. Pointing to Jones's agreement with the community does not help us here, since at best it simply describes the result of Jones's training and says nothing about how it is that Jones has come to that agreement. When we frame the matter in this way, we bring out the fact that agreement is itself liable to be explained as obtaining on the basis of some other fact or set of facts, specifically, those constituting Jones's learning as such. [34] Interestingly, despite Kripke's occasional allusions to learning, it is not clear that the phenomenon plays an important role in his skeptical solution. On the contrary, it would seem that it does not. There are few references to learning in Kripke's exposition of the skeptical solution, and when we do run across them they seem either tangential to the main argument (e.g., WRPL 82 n 71) or illustrative of, or otherwise subordinate to, the notion of public checkability (WRPL 89-90; 99-100). I believe Fogelin is right to assert that Kripke relies "almost exclusively" on the public check argument (Fogelin 244, compare also 183), and gives little consideration to the matter of learning. But I think this is understandable, given that the consequences of an explicit acknowledgement that agreement is learned would introduce into the equation a set of facts that can only serve to undermine the claim that agreement is the brute fact of the matter. [35] If agreement is not the brute fact of the matter, then what is? The observational model of learning I suggested in my earlier paper was meant to hypothesize what that other fact or set of facts (e.g., processes, mechanisms, conditions, etc.) could be. I chose the case of a simple, publicly observable skill that could be reproduced through the learner's observation, selection of relevant features, and construction of an internal model representing the relevant requirements. Note that this model was intended only to address the limited question of how the reproduction of behavioral regularities characteristic of a given population could in principle be explained in terms of the relevant constituent factors. The simple observation/selection- internalization/construction process sketched would have to be modified to explain a broader range of learning situations, though I think the generalities of the account would still hold. [36] I will now look briefly at some of the implications of the model. In doing so, I hope to clear up some misconceptions Huen seems to hold regarding the picture I sketched. For it seems that where I outlined a simple process of behavioral reproduction by way of a cognitive process of pattern recognition and modeling, Huen instead sees a process involving the direct transmission of a fixed object that is the same for all who can be said to have internalized it (Huen b 2.3). Huen's reading serves to mislead him on a number of points regarding more local issues. Rather than attempting to address each of these latter, I will restrict myself in the following sections to elaborating the basic features of the model, with extrapolation to subsidiary points where appropriate. 2.3. Construction is not Direct Transmission [37] The first and most fundamental point I would like to make is that the model I sketched is not a direct transmission model. Such a model would rely on two important assumptions: the first is that the learner is a more or less passive recipient of whatever it is that is transmitted, and the second (which is closely related to the first) is that the transmitted element is the same for all who can be said to receive it. The direct transmission model would, in other words, posit some item "out there" that could be gotten directly "in here" for those who would learn it. [38] By contrast, the process I adumbrated is one in which the learner internalizes the normative requirement through an interplay of observation, feature selection, and engagement of his or her conceptual repertoire. The result is a kind of map or template representing relevant aspects of the observed example. By this reading, "internalization" is a gloss on a process by which the observer's cognitive system constructs, through a complex interaction of perceptual input and an existing conceptual repertoire, a model of the example to be reproduced. Put simply, an external object isn't (somehow) gotten inside the learner; rather, an internal map is constructed on the basis of a weighted interaction of external features and the relevant cognitive mechanisms. It is by virtue of this construction that we can say that the relevant salient features of the example have been internalized, with the following proviso: "Internal" should be recognized as a perhaps unfortunate but ultimately unavoidable figure for describing one aspect of information states in the brain in relation to which -- and on the basis of which -- behaviors are organized. [39] What is important here is that the normative requirement embodied in, e.g., a performance is not transmitted but rather is constructed or hypothesized by virtue of its being modeled by the learner. If this is so, then -- again, the main point of my earlier paper -- agreement looks less like a brute fact than a phenomenon that not only should be explained, but can be explained, in terms of other facts. [40] Several consequences follow from a model based on construction rather than direct transmission. First, we cannot think of the normative requirement as a Platonist object of some sort. Second, the learner's internal representation need not be thought of as consisting of an inventory of necessary and sufficient conditions. Third, each individual's grasp of the normative requirement will be marked by variation(s) peculiar to that individual -- there will be a relation of commensurability rather than identity among different people's understanding of the normative requirement, in other words. Fourth, the process of learning described may be thought of as allowing a variety of habituation. I will briefly look at each of these points in turn. 2.4. No Transcendent Objects [41] Perhaps the most important consequence of positing a constructivist process over one of direct transmission is that there is no need to appeal to some kind of Platonist fixed object or "entity of [a] pre-fixed nature" (Huen b 2.3), since there is, strictly speaking, no thing that is transmitted. The closest candidate we can point to in this regard is the publicly observable example presenting a particular gestalt some features of which will take on a certain relevance or salience in relation to the task at hand and thus to the normative requirement involved. But this isn't very close at all, and there is no reason to speak of transcendent objects when discussing the salient features of publicly accessible examples. [42] It is possible that Huen's confusion of the normative requirement with a fixed object of some sort stems from a misunderstanding of what in my earlier paper I termed a norm's transcendence condition (Barbiero 1996 section 3). This is simply a way of expressing the fact that a performance may embody a requirement (which I described as its "normative content"), which will carry a binding force transcending any specific enactment or embodiment. This transcendence condition is just implicit in the notion of a norm: norms impose a requirement on us, and aren't just anything we say they are. To this extent normative requirements are not exhausted in the performances we undertake in applying them, and consequently they can be said to provide the standards in terms of which the judgments informing those performances may themselves be judged. In this sense, the norm's requirement transcends any particular performance or judgment in relation to which it is binding, and this, as Wright points out in his discussion of Kripke's skeptical solution, is true even of consensual judgments (Wright 244). There is a second, related sense in which a norm is transcendent, and that is in regard to its binding of intentions in appropriate situations in the future. This, as Kripke points out (WRPL 37), is one important way in which a norm differs from a disposition. In sum, "transcendence" here is being used in a very limited way to mean that a norm's requirements are binding beyond any given (present) instantiation. No conclusions attributing a fixed, object-like status to the normative requirement should be drawn from this. 2.5. Necessary and Sufficient Conditions are Not Necessary [43] It does not necessarily follow from the above model that the learner's mental representation of the normative requirement will consist in an inventory of necessary and sufficient conditions. On the contrary, there is no reason this representation cannot consist in a schematic characterization of weighted features emerging at the requisite level of generality (given, as well, the appropriate external constraints on feature salience). To hypothesize this type of representation is not necessarily to commit oneself to strong generativity, i.e., to the position that representations consist in recursive descriptions of all appropriate applications of the normative requirement, in all cases. Instead, we might say that in the case of learning under consideration, the normative requirement is represented to a first approximation as a more or less coarse grained generalization emerging on the basis of the appropriate feature detection and categorization. In fact it is this variety of representational makeup that we might expect of the kind of pattern-based learning capacity presented. 2.6. Individual Variation [44] Just as there is no question of an abstract object of some sort that somehow must be gotten into the heads of all those who can be said to share a given behavioral pattern, there is no question of those behaviors being produced by an internal factor that is somehow the same for all (cf. Huen b 2.3). This is true whether we take a nativist position or a position deriving from empiricist associationism. [45] Consider that the constructivist model takes learning to be a process arising from the interaction of external and internal constraints. As I suggested in my earlier paper, the former can be thought of as consisting in an organized stimulus field, or environmental gestalt. The latter would consist in the learner's cognitive mechanisms, which we can assume will include the capacities, generalized or task-specific as the case may be, to select ambient information and categorize it according to a given conceptual repertoire. Whether we try to account for these capacities by claiming them as innate or as ontogenetically acquired, we must grant that they will be marked by a degree of variation by individual. The reason for this is simple. [46] If, like Chomsky, we conceive of such internal constraints as consisting in a genetically transmitted, innate device of some sort, we find ourselves having to employ what has been called the population mode of thinking (Lieberman 1984, 14). This requires us to acknowledge the fact of variations among individuals because, as Lieberman has pointed out, genetic variations are always present in populations and any genetically-transmitted capacities, cognitive or otherwise, would have to be assumed to be sensitive to such variations (Lieberman 1991, 130-134). (Chomsky, interestingly, appears to deny that such variation is a factor in determining different individuals' language capacities.) If on the other hand we are to claim that internal constraints are the result of experience, i.e., are structured by virtue of the cognitive mechanism's interactions with and adaptations to its environment, we then find ourselves confronted by the fact that, no two individuals having the same experiences, their cognitive organizations will differ accordingly. In either case, it is untenable to hold that the internal constraints working to produce behaviors will be the same for different individuals, even if the behaviors in question are appreciably similar to all outward appearances. [47] It thus seems to me that the strongest claim we can make is that those who can be said to internalize, in the sense explained above, the relevant features of a common environment most likely will be marked by a certain commensurability of cognitive content regarding those features. (By commensurability I simply mean a rough comparability or similarity.) This likelihood is underscored when we take into account the fact that through what Putnam has termed the linguistic division of labor (Putnam 1991, 22), groups of speakers have access to experts and information that may serve to introduce a measure of rough standardization into the content of their conceptual repertoires. Still, nothing stronger than commensurability can be assumed, and even then the claim most likely would have to be weakened the finer-grained a description we would want from different people. 2.7. Habituation [48] The simple account of learning that I have been discussing hypothesizes mechanisms and processes that can be seen to support adaptive habituation based on cognitive reorganization. (For present purposes, "habit" might be defined as an acquired, patterned response afforded by the plasticity of the organism's cognitive endowment.) This has consequences that put the skeptical paradox in an interesting light, since it is often on the basis of habituation that one can apply a skill or conform to a rule without need for deliberation. Fogelin, for one, has proposed habituation (without actually calling it such) as a solution to Kripke's skeptical paradox (Fogelin 185), and it is a solution that I find attractive, provided one makes allowance for the fact that there would seem to be a discretionary moment in which one must come to the recognition that the rule applies in the given situation.*4* [49] I would add that thinking in terms of habituation has the advantage of allowing us to account for the changes in response - - often generational -- that characterize populations over given periods of time. It seems to me that if we consider learning to involve a process of habituation, we necessarily assume a degree of cognitive plasticity that in principle is sufficient to allow for change. If we think of change as a kind of adaptation through which we (and other animals for that matter, though certainly to differing degrees) exploit the plasticity that characterizes our cognitive mechanisms, the question would seem to answer itself, since such plasticity is the prerequisite for the creative adaptivity involved in change. A kind of incremental change might also accrue simply as an incidental product of within-population variations. While it is true that completely hard-wired capacities, if any, presumably would not themselves change, the picture is considerably complicated by the likelihood that there is some interaction between such capacities and experience-sensitive capacities. [50] It seems to me that two qualifications are in order. First, habituation as I am presenting it here should not be equated with dispositionalism. Second, there are limits to the types of human phenomena that can be explained in terms of habituation. [51] I mention the first qualification because it appears that Huen has -- incorrectly -- interpreted the learning model presented as dispositional (see Huen b 2.3). Although I quite explicitly reject a dispositional response to the skeptical paradox (Barbiero 1996 section 5.4), some confusion may have arisen due to my characterization of normative content as playing a causal role in the production of behavior (Barbiero 1996 section 4.1). The claim comes down to this: normative constraints cause behavior to the extent that mental states cause behavior on the basis of what they represent. This is a statement about the role representational states play in the production of behavior generally, and not a reduction of normative considerations to dispositional determinations. To hold, as I do, that behaviorally speaking, reasons -- normative or otherwise -- are causes, is not to hold that given a certain stimulus, one will invariably (or nearly so) produce a given response. [52] The second qualification is a bit more complicated. It would seem that as an explanatory factor, habituation has its limits. The range of behaviors that can be attributed to habituation is not, in other words, unlimited. It is not clear, for example, that we can speak of habituation in relation to language behaviors generally, although it has been hypothesized that syntactic rules make use of the sort of automatized neural subroutines associated with the effects of habituation (Lieberman 1984). Certainly, the extent of training and the role it plays in language acquisition is not settled, though few at this point would likely accept the behaviorist account of language learning via reinforcement by a speech community -- although there is still debate over whether the stimulus is in fact as impoverished as is claimed by some. Nor does it seem likely that habituation by itself can explain action in situations that call for creative responses rich in novelty -- but even here it may be reasonable to suppose that habits of various sorts would provide the substrate from which novel responses could be staged. [53] But it does seem to me that even if habituation cannot account for everything, it does provide a plausible explanation for the mastery of a skill like arithmetic. Here we might raise the point that there are significant differences between what it takes to possess a concept and what it takes to master a skill, and that it may not be appropriate to consider skill mastery as a subset of concept possession, as Kripke's skeptical solution apparently does. For although concept acquisition reasonably can be expected to involve cognitive plasticity and reorganization, it is not clear that the plasticity and reorganization involved will be comparable to that involved in habituation. Of course, if we are only concerned with whether or not people's responses agree, then this would not appear to be an issue. But that I think is indicative of nothing so much as the limitations of an approach concerned only with responses and not with the deeper explanation of how those responses come to be. Daniel Barbiero Notes *1*. Let me say a word about why in my earlier paper I referred to agreements in the plural. As I read it, Kripke's definition of a form of life as a set of responses in which we agree seems to imply a bundle of separate but related conventions, or agreements. By this reading, a form of life is a domain or field characterized by behavioral regularities pertaining to the appropriate goals, methods, habits, etc. The form of life for Kripke, in other words, seems to represent an aggregation of behavioral regularities grounded in community agreement. *2*. Here, again, the phrase "depends on," to which Huen raises a curious objection (Huen b 2.2). Huen somehow -- showing neither evidence nor the path of reasoning that would render it a necessary conclusion -- interprets my paraphrase of Kripke's position ("the correct interpretation of a rule...depends on conformity to the appropriate behavioral regularities") as meaning that such conformity is "the necessary and sufficient condition of following or defining a rule." But "depends on" simply means "is determined by or in relation to." I deliberately chose this wording in order to be faithful to two points I see in Kripke's account. The first is the nuance inherent in Kripke's observation that we display "rough uniformities" of response (WRPL 97) in that we "generally" agree (WRPL 97) in "enough particular cases" (WRPL 106), and his consequent implicit allowance for a tolerable amount of deviation across responses. It seems to me that such allowance disappears when we start talking about necessary and sufficient conditions. The second point I wished to conserve with this wording is the skeptical solution's rejection of the autonomy of rules. While it is not always clear that Kripke accepts the psychological reality of rules, it does seem to be a major point of the essay that there cannot be autonomous, i.e., agreement-independent, rules. *3*. The model of reproduction sketched in the earlier paper focused on the external aspects of the process. In the present paper I will have more to say about some of the internal aspects. *4*. For this reason, I find it potentially misleading to talk of the "non-interpretive" application of a rule. Such a phrase would seem to lend itself to a confusion between the pragmatics of rule-following and the content of the rule itself. Additional Bibliography Barbiero, Daniel. "Chomsky v. Kripke, Round Two: Methodological Collectivism and Explanatory Adequacy" (1996). Chomsky, Noam. "Linguistics and Adjacent Fields: A Personal View." In Kasher, pp. 3-25. Chomsky, Noam. "Linguistics and Cognitive Science: Problems and Mysteries." In Kasher, pp. 26-53. Fogelin, Robert J. WITTGENSTEIN, Second Edition (London: Routledge, 1987). Huen, Kenny. "Clarifying the Skeptical Solution in Kripke's Wittgenstein" (1996). Internal cites to Huen b. Kasher, Asa. THE CHOMSKYAN TURN (Cambridge, MA: Blackwell, 1991). Internal cites to Kasher. Lieberman, Philip. THE BIOLOGY AND EVOLUTION OF LANGUAGE (Cambridge, MA: Harvard University Press, 1984). Lieberman, Philip. UNIQUELY HUMAN: THE EVOLUTION OF SPEECH, THOUGHT, AND SELFLESS BEHAVIOR (Cambridge, MA: Harvard University Press, 1991). Putnam, Hilary. REPRESENTATION AND REALITY (Cambridge, MA: MIT Press, 1991). Pylyshyn, Zenon. "Rules and Representations: Chomsky and Representational Realism." In Kasher, pp. 231-251. Stern, David G. WITTGENSTEIN ON MIND AND LANGUAGE (Oxford: Oxford University Press, 1995).